GLP-1 vs. bariatric surgery costs: What a long-term study reveals for healthcare purchasers

As healthcare economists and policy experts strive to assess the value of covering GLP-1 drugs, one key step is to compare the potential ROI with that of established weight-loss treatments.

To that end, Israeli researchers recently published a study evaluating the long-term costs of GLP-1 receptor agonists (GLP-1RAs) relative to bariatric surgery for patients with obesity and diabetes. The authors excluded the cost of the interventions, focusing instead on downstream healthcare utilization.

As published July 1, 2026, in JAMA Network Open, the study linked bariatric metabolic surgery (BMS) with lower long-term healthcare costs, relative to GLP-1RAs. The authors found an adjusted difference of $109 in monthly healthcare utilization costs per patient.

“The advantage associated with BMS vs pharmacotherapy in terms of the cost of healthcare utilization we report here is in accordance with the greater clinical effectiveness of BMS compared with GLP-1RAs previously observed,” the authors wrote, citing prior findings of superior impacts on mortality and heart failure hospitalization rates among patients with congestive heart failure.

One caveat is that the study examined primarily first-generation GLP-1 agents. Widespread inclusion of later iterations such as semaglutide and tirzepatide products might alter the equation. Despite their high price tags, GLP-1s also represent a more accessible treatment than BMS for many people.

In this Q&A, the study’s corresponding author, Yael Wolff Sagy, PhD, provides additional insight on the implications of the study findings. She’s an epidemiologist with Clalit Health Services, Israel’s largest state-mandated health maintenance organization and healthcare provider.

Note: Answers have been edited for length and clarity.

Were there any insights from your study aside from the main conclusion that stood out to you?

Beyond the overall finding that patients treated with GLP-1 receptor agonists had higher downstream healthcare utilization costs than those who underwent bariatric-metabolic surgery, what struck us was the timing of those differences.

Most of the cost divergence emerged early. Roughly three-quarters of the total long-term difference accumulated within the first four years of follow-up. At the same time, bariatric surgery produced much greater early reductions in both BMI and HbA1c, whereas GLP-1 therapy resulted in more modest but sustained improvements. This temporal relationship suggests that the speed and magnitude of early metabolic improvement may be a key driver of later healthcare utilization.

Is there a way to estimate how much these findings might change if incorporating updated GLP-1s like semaglutide or tirzepatide? Could greater weight loss change the hospitalization trajectory?

I want to be clear that [the following] is personal judgment extrapolating from the results, not a finding from the study itself.

That said, I think it’s reasonable to expect the gap in downstream healthcare costs to narrow with newer agents. Semaglutide and tirzepatide generally produce greater weight loss and stronger glycemic improvements than the agents used by most patients in our cohort. If hospitalization risk is partly mediated through obesity-related complications and metabolic control, stronger weight loss should translate into lower hospitalization costs and lower reliance on additional medications.

However, I would not assume that newer drugs would eliminate the advantage we observed for surgery. Bariatric surgery still induced the most rapid and profound early weight loss and diabetes improvement. My expectation is that the cost curves move closer together, particularly in the first few years, but whether they converge fully is an open empirical question that needs its own cohort with real follow-up on these newer agents, not an extrapolation from ours.

How transferable are these findings to the U.S. healthcare system, where insurance coverage is fragmented? Which findings do you think are universal?

The exact magnitude of the findings may differ in the United States, but I believe that the underlying mechanisms are broadly applicable.

Our study was conducted within Israel’s universal healthcare system, where both bariatric surgery and GLP-1 therapy are publicly funded within the same integrated framework for persons with obesity and diabetes. That allowed us to observe long-term healthcare utilization without many of the coverage interruptions common in fragmented insurance environments.

Nevertheless, the principles are likely universal: Greater weight loss and better metabolic control tend to improve obesity-related morbidity and reduce future healthcare utilization. I’d therefore expect the direction of the findings to be transferable even where the precise numbers are not.

What additional data elements or methodological changes could contribute to a decision-ready economic model indicating whether expanded GLP-1 coverage creates a positive ROI?

A decision-ready model would require four key additions. First, it would need to include all treatment-related costs, including drug acquisition costs, rebates, monitoring, physician visits and adverse-event management.

Second, it should capture a broader range of outcomes, such as cardiovascular events, kidney disease progression, diabetes remission, mortality and quality of life, since many important benefits may emerge years after treatment initiation.

Third, persistence [of use] should be modeled explicitly. The economics of GLP-1 therapy are highly dependent on how long patients remain on treatment. A payer could reach very different conclusions if most patients discontinue within a year versus remaining on therapy for many years.

Finally, the model should account for patient heterogeneity. Not all patients are equally likely to benefit. Individuals at particularly high baseline risk of hospitalization or cardiometabolic complications may generate greater downstream savings than lower-risk patients. For that reason, I believe future work should move beyond average ROI estimates and focus on identifying which patients are most likely to generate positive long-term clinical and economic returns.

Ultimately, the ideal analysis would combine long-term clinical outcomes, claims data, pharmacy expenditures and treatment persistence in a large, real-world population and [would] project costs and savings over a lifetime horizon. Only then can we answer the question most relevant to payers: Does the reduction in future healthcare spending outweigh the cost of providing GLP-1 therapy, and for which patients?

If you had access to thorough longitudinal claims and clinical data, what variables would you use to predict which patients will generate the largest downstream savings from GLP-1 therapy? What costs should be included in such a calculation?

I’d begin with a patient’s sex and age, along [with] clinical characteristics and risk factors, including baseline BMI, HbA1c, hypertension, lipid profile, and cardiovascular and kidney disease. Prior healthcare utilization would also be highly informative.

Perhaps most importantly, I’d include measures of early treatment response. The amount of weight loss and HbA1c reduction achieved in the first three to six months may be among the strongest predictors of long-term clinical and economic benefit.